On Tuesday, after a less-than-stellar debate performance, Donald Trump returned to using one of his favorite measurements to mask his missteps on Monday night — the polls. Except that the post-debate "polls" the GOP nominee and campaign kept touting that he won on social media, at his rally in Florida and in press releases aren't polls at all. They're essentially unscientific Internet popularity contests, are not weighted as to what the electorate will actually look like, and have no predictive value. In fact, if you're worried about voter fraud, in many of these surveys people can vote multiple times and they can easily be rigged by Internet bots.
Many of the results he cited came from very Trump-friendly sites, such as Breitbart News and the Drudge Report. Some of these polls were shared by Trump supporters on reddit, encouraging people to go vote. These were not selective samples with any merit, and in no way could they accurately measure whether the more than 81 million people who tuned in for the debate thought Trump did better than Hillary Clinton. Even though Fox News repeatedly touted the same results Trump trumpeted on air and online, Business Insider reported that the network's polling director, Dana Blanton, warned that such polls "do not meet our editorial standards":
"As most of the publications themselves clearly state, the sample obviously can't be representative of the electorate because they only reflect the views of those Internet users who have chosen to participate. ... Another problem — we know some campaigns/groups of supporters encourage people to vote in online polls and flood the results."
Huffington Post polling director Ariel Edwards-Levy explained the vast difference between these polls Trump loves so much and ones that are actually predictive and useful:
"Scientific polls can be conducted by phone, via online panels or some other way, and use a mix of sampling and weighting to make their numbers representative of the larger population whose opinions they're measuring ― whether that's all adult Americans, or just likely voters. Recent changes in technology have complicated that process, but the underlying principle remains basically the same.
"That's why, even if a scientific poll reaches relatively few people, it can accurately depict the opinions of a much larger group.
"In contrast, reader polls, like those Trump cites in the tweet below, make no such attempt to weight their responses or to represent anything beyond the number of people who happen to have clicked on them."
Republican pollster Glen Bolger said it was unfortunate that these post-debate surveys, intended to drive traffic, were even conducted because they muddied the water for legitimate polls.
"It's a disservice by the media outlets that do them, and there's nothing scientific, nothing rigorous about them," Bolger said. "It's whoever goes on the website and wants to take the poll. And obviously a bunch of them are going to be self-selected toward the partisanship of their readership. There's just nothing good in them, there's just no point to them."
In fact, actual scientific polls have shown that Clinton was the clear winner on Monday night. A CNN poll just after the debate showed 62 percent of voters thought Clinton had the best performance, while just 27 percent said Trump did. The sample from the call-back survey did lean a bit too Democratic (the pollster didn't have time with such a quick turnaround to weight the sample), but even a minor adjustment would still equal a clear victory for the Democratic nominee. A Politico/Morning Consult poll also gave Clinton the win, 49 percent to 26 percent. Another survey, from the Democratic-leaning Public Policy Polling, also found Clinton won by 11 points, 51 percent to 40 percent. A two-day poll from Echelon Insights found that 48 percent of registered voters thought Clinton won the debate, while just 22 percent thought Trump did.
It will take several days for the impact of Monday night's debate to be fully shown in polls, Bolger said, and data on Friday morning that survey people over several days would provide the best measurement. But even then, in such a fast-changing campaign, it might only be a brief flashpoint before the next presidential debate on Oct. 9.
But for now, discount these "flash" polls and wait for more reputable numbers to come out. Remember, in 2012 these same flash surveys all found that former Texas Rep. Ron Paul won the GOP debates by wide margins. But, no, there was no President Paul. Copyright 2019 NPR. To see more, visit https://www.npr.org.
(2016-09-28 04:00:00 UTC):
A previous version of this story incorrectly said that results of a CNN poll showed 27 percent of voters thought Hillary Clinton won the debate. The 27 percent actually refers to Donald Trump.